Documentos de Académico
Documentos de Profesional
Documentos de Cultura
gst Gumundsson
Faculty
Faculty of
of Civil
Civil and
and Environmental
Environmental Engineering
Engineering
University
University of
of Iceland
Iceland
2013
2013
gst Gumundsson
Advisors
Dr. Sigurur Magns Gararsson
Dr. Gunnar Guni Tmasson
Mr. Andri Gunnarsson
Faculty Representative
Mrs. Hrn Hrafnsdttir
Bibliographic information:
gst Gumundsson, 2013, Model Investigations of a Juvenile Fish Bypass System at
Urriafoss HEP , M.Sc. thesis, Faculty of Civil and Environmental Engineering, University of Iceland.
Abstract
The flow characteristics of an intake pond and a Surface Flow Outlet (SFO) type juvenile
bypass system were investigated using a 3D computational fluid dynamics (CFD) model
and a 1:40 scale physical model. In addition the flow conditions inside SFO conveyance
channel were evaluated in a separate CFD model. The intake pond and juvenile bypass
system are a part of proposed hydroelectric power plant in the Lower jrs River in
Southern Iceland which is located in the migratory pathway of the North Atlantic Salmon.
The approach flow conditions in the intake pond and flow conditions inside the SFO conveyance channel were investigated to ensure safe and timely passage of juvenile salmon
through the system. Three cases of operational conditions were investigated for the approach flow conditions in the reservoir while the conveyance channel was evaluated with
regard to four different discharge rates. The numerical model used was the free surface
model of ANSYS-CFX software with a standard - turbulence model. The numerical
results were compared and validated with the physical model results by comparison of
velocity profiles from an Acoustic Doppler Velocimeter (ADV), particle tests, dye tests
and visual observations from the physical model study. In general the SFO design and
approach flow conditions prove to be effective in providing a safe and timely passage of
juvenile salmon through the system. There is a good agreement between the results of the
physical and the numerical models closest to the intake and spillway structures. Further
upstream the results between the physical and numerical model started to differ due to
minor inconsistencies between numerical and physical model topography.
tdrttur
Arennslisskilyri inntakslni virkjunar og rennslisskilyri seiafleytu voru rannsku me tvennskonar lknum, annars vegar rvu tlulegu lkani og hinsvegar straumfrilegu lkani. Inntakslni og seiafleytan eru hluti af fyrirhugari Urriafossvirkjun
Neri jrs. Virkjunin verur stasett miri gngulei laxfisks, til a lgmarka hrif
hennar laxastofn rinnar verur seiafleyta samt laxastigum bygg. Arennslisskilyri inntakslni og rennslisskilyri innan seiafleytunnar voru rannsku til a tryggja
rugga og skjta fer seia gegnum hrifasvi virkjunarinnar. rj rekstrartilfelli
voru rannsku fyrir arennslisskilyrin og fjgur tilfelli me mismiklu rennsli innan
seiafleytunnar. Tlulegu lknin voru bygg upp me hugbnainum ANSYS CFX ar
sem nota var lkan fyrir rennsli vi frjlst yfirbor me hefbundnu - iustreymislkani. Niurstur tlulega lkansins voru bornar saman vi mlingar r straumfrilegu
lkani, ar meal hraasni r Acoustic Doppler Velocimeter (ADV), rennslihegun r
agnaprfi ofl. heildina liti er hnnun seiafleytunnar og arennslisskilyri lkleg til a
laa seiin a seiafleytunni og skila seiunum me skjtum og ruggum htti gegnum
kerfi. gtt samrmi er milli niurstaa tlulegs lkans og straumfrilegs lkans
nst inntaki virkjunar og seiafleytu en lengra fr mannvirkjunum var samanburur
niurstum ekki eins gur vegna ltilshttar samrmis landslagi milli lkana.
vi
Contents
List of figures
ix
List of tables
xiii
Acknowledgments
1 Introduction
1.1 General Introduction . . . . . . . . . . . . . . . . . . . . .
1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4 Co-ordination Groups . . . . . . . . . . . . . . . . . . . . .
1.5 Literature Review . . . . . . . . . . . . . . . . . . . . . . .
1.5.1 Behavior of Migrating Juvenile Salmon . . . . . . .
1.5.2 Development of Downstream Fish Passage Systems .
1.5.3 Numerical Models . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
3
5
6
7
8
8
9
12
2 Methods
2.1 Theory . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.1 Governing Equations . . . . . . . . . . . . . .
2.1.2 Turbulence Models . . . . . . . . . . . . . . .
2.1.3 Volume of Fluid Method . . . . . . . . . . . .
2.1.4 Numerical Solver . . . . . . . . . . . . . . . .
2.2 SFO Conceptual Framework . . . . . . . . . . . . . .
2.3 SFO Design Guidelines and Criteria . . . . . . . . . .
2.3.1 Design criteria upstream of SFO entrance . . .
2.3.2 Design criteria in SFO and conveyance channel
2.3.3 Design criteria at outfall . . . . . . . . . . . .
2.3.4 Summary of design criteria . . . . . . . . . . .
2.4 Numerical Model . . . . . . . . . . . . . . . . . . . .
2.4.1 Geometry . . . . . . . . . . . . . . . . . . . .
2.4.2 Mesh . . . . . . . . . . . . . . . . . . . . . .
2.4.3 Solver set up . . . . . . . . . . . . . . . . . .
2.4.4 Execution . . . . . . . . . . . . . . . . . . . .
2.5 Physical Model . . . . . . . . . . . . . . . . . . . . .
2.5.1 Velocity Measurements . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
15
15
15
16
16
18
18
20
20
20
21
21
22
22
25
27
28
29
32
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
vii
Contents
2.6
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
32
32
33
33
33
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
35
35
35
40
43
45
52
58
65
66
67
69
69
75
76
77
4 Conclusion
79
References
81
viii
List of figures
1.1
1.2
1.3
2.1
18
2.2
23
2.3
23
2.4
24
2.5
24
2.6
Ogrid mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
2.7
25
2.8
26
2.9
26
26
29
31
31
32
ix
LIST OF FIGURES
3.1
36
3.2
36
3.3
37
3.4
37
3.5
38
3.6
38
3.7
39
3.8
39
3.9
41
3.10 Top view and longitudinal section of the intake and SFO structure . . . .
42
44
44
48
49
49
50
50
51
54
55
55
56
LIST OF FIGURES
3.23 Velocity contours numerical model Case B . . . . . . . . . . . . . . . . .
56
57
58
61
62
62
63
63
64
64
3.33 Dye drawn towards SFO entrances 2 and 3 during a dye test . . . . . . .
65
66
67
69
70
71
72
74
75
3.42 Water levels and velocity vectors at longitudinal sections in the SFO . . .
76
78
78
xi
List of tables
2.1
Computational time . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
2.2
33
3.1
43
3.2
65
3.3
Calculations of k2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
73
3.4
77
xiii
Acknowledgments
I would like to thank my advisors Dr. Sigurur Magns Gararsson, Dr. Gunnar Guni
Tmasson and Mr. Andri Gunnarsson. I feel very privileged for the opportunity to
take part in the Urriafoss HEP physical model study and truly appreciate the trust my
advisors showed me. Their support and professional criticism has been of great value
during the course of the study.
I would like to thank Dr. Helgi Jhannesson at Landsvirkjun for his general interest
towards my study and for the collaboration and support throughout the study.
The designers at Verks and Mannvit, Ms. lf Rs Kradttir, Mr. orbergur Leifsson
and Mr. Einar Jlusson get my thanks for the collaboration during the study.
My thanks go to Dr. Sigurur Gujnsson, the director of the Institute of Freshwater
Fisheries for his help with the fisheries literature.
I would also like to thank the craftsmen at the Icelandic Maritime Administration for their
great help and collaboration during the building of the physical model.
I would like to thank Mr. Gsli Steinn Ptursson, a laboratory researcher and colleague
during the physical model study, for his help in data acquisition, handling and presentation
and for the great companionship throughout the study.
1 Introduction
1.1 General Introduction
Of the river systems in Iceland the jrs and Tungn river system has the greatest
hydropower potential providing 27% of all economically viable hydroelectric energy in
the country. The drainage area of the system is 7530 km2 were 1200 km2 are covered by
glaciers characterising the river mainly as a surface runoff river with glacial and spring
influences.
The development of the jrs and Tungn river system started with commission of the
Brfell hydroelectric project (HEP) in 1969. Four additional power plants and extensive
diversion systems in the uppermost part of the river system have been commissioned since
then making the river highly regulated. In late 2013 the fifth power plant Barhls HEP
will start operation utilizing the head between Hrauneyjafoss HEP and Sultartangi HEP.
Landsvirkjun is preparing construction of three hydroelectric power plants in the Lower
jrs River, Hvammur HEP 81 MW, Holt HEP 53 MW and Urriafoss HEP 128 MW,
shown in Figure 1.1. Because of the highly regulated characteristics of the Lower jrs
River no additional storage is needed for the projects.
The project furthest downstream, Urriafoss HEP, utilizes the head difference between
elevations of 50 m.a.s.l. and 9.4 m.a.s.l. The design discharge at Urriafoss is 370 m3 /s
with energy generating capacity of about 980 GWh/year.
The layout of the Urriafoss HEP, shown in Figure 1.2, is in sense typical for a run of the
river hydro electric plant with the exception of additional structures to aid the migration
of salmon through the project. The additional structures to aid salmon migration are
a Surface Flow Outlet (SFO) type juvenile fish bypass system, upstream fishway and
a mandatory release structure to provide constant 10 m3 /s discharge in to the riverbed
downstream of the dam. The powerhouse will be equipped with two minimum gap Kaplan
turbines which are considered to be fish friendly (Gudjonsson & Johannesson, 2009).
1 Introduction
Figure 1.1: Overview of the Lower jrs River hydroelectric projects (Landsvirkjun,
2010).
1.2 Motivation
A small 9 km2 intake pond, Heiarln reservoir, is formed at Heiartangi point with 7.5 km
long dykes lying along the west bank of the river.
Normal Reservoir Water Level (NRWL) is 50 m
a.s.l. but during the design flood event the High
Reservoir Water Level (HRWL) is 51.5 m a.s.l.
The spillway structure consists of three radial
gates with a roller bucket type energy dissipator.
The SFO is located above the turbine intake as
shown in Figure 1.3 and has four 5.95 m wide
entrances, each with a smooth rounded crest at
an elevation of 49.1 m a.s.l. providing an estimated discharge of 40 m3 /s at NRWL. From the
crest the water from the four entrances is united
in a single sideway channel and routed through a
4.5 m wide concrete channel 90 m downstream
of the dam to the original riverbed. The elevation
difference between the entrance of the SFO and
outfall is 8 m. The SFO is designed to transport Figure 1.3: Schematic longitudinal view
the uppermost 1 m layer of water and to create of SFO and power intake structures.
favourable conditions to attract juvenile salmon
towards the SFO entrance.
1.2 Motivation
In recent years environmental issues associated with the construction and operation of
hydroelectric power plants have been receiving more and more attention. With the proposed
Lower jrs River hydroelectric projects outmigration of juvenile salmon becomes an
factor in the design not encountered before in Icelandic hydroelectric projects. The limited experience provides ground for expanding the local knowledge base by seeking out
global successes and factors necessary for a successful design. By identification of factors
influencing the response of juvenile salmon to unnatural flow features associated with
the operation of hydroelectric power plants necessary measures can be made to ensure
successful operation and thus limit the overall impact of the project.
In the environmental impact assessment for the Lower jrs River projects from 2003
mitigation measures are listed that are intended to sustain the natural salmon stock of the
jrs River. The mitigation measures consist of a mandatory release structure intended to
maintain a minimum flow of 10 m3 /s (environmental flow) to the original riverbed downstream of the dam, a upstream fishway and a juvenile fish bypass structure for downstream
1 Introduction
migration of juvenile salmon.
The study presented in this thesis is part of the physical model study of the Lower jrs
River projects Hvammur HEP and Urriafoss HEP carried out at the hydraulic laboratories
of the Icelandic Marine Administration in Kpavogur, Iceland. The project is a collaboration of Landsvirkjun, the University of Iceland, the Reykjavk University, the Icelandic
Marine Administration and the designers Mannvit and Verks consulting engineers. The
first phase of the project, the Hvammur HEP physical model, was built and tested through
out the year 2011. The model was the first physical model study to be conducted in Iceland in over twenty years. The second phase, the Urriafoss HEP model test, was built and
tested from February to August 2012. As part of the project three master thesis are carried
out at the University of Iceland. From first phase Andri Gunnarsson carried out a study
of the spillway at Hvammur (Gunnnarsson, 2012), from that followed a full technical
report (Tomasson et al., 2012) and a conference article (Gunnarsson et al., 2012). From
the second phase two master thesis are carried out, the study presented in this thesis and
a master thesis on the roller bucket type energy dissipator carried out by Gsli Steinn Ptursson. In addition from the second phase a full technical report on the physical model
study (Tomasson et al., 2012) and a conference article (Gudmundsson et al., 2012) were
released.
As a further expansion of the physical model study a numerical model of the juvenile
bypass system and pond at Urriafoss HEP was built. The numerical model and physical
model are intended to compliment each other with the numerical model shedding light
on delicate features in the approach flow and other features which are hard to quantify in
the physical model. The results from both models represent a realistic indication of flow
conditions expected on a prototype scale.
1.3 Objective
The objective of this study is to evaluate the design of the proposed Surface Flow Outlet
(SFO) type juvenile fish bypass system at Urriafoss HEP using two models, a physical
model and a numerical model. The models are compared and validated by various methods including comparison of velocity profiles and general flow behaviour from particle
tests from the physical model study. The validated numerical model is in addition used to
look into features hard or impossible to measure in the physical model which are important for evaluation of the design with regard to the juvenile salmon. The evaluation is
mainly focused on approach flow conditions upstream of the SFO entrances and flow
conditions inside the SFO conveyance channel where conditions are evaluated to ensure
safe and timely passage of juvenile salmon.
1 Introduction
1 Introduction
River systems extensive research on the behaviour of migratory fish close to dams and
design of more efficient DMS has been undertaken. This extensive research has generated
considerable but often inconclusive results as to factors influencing migrant guidance and
passage (Goodwin, 2003). Early bypass systems often achieved only limited and variable
success at considerable cost because migrants either could not locate the bypass entrance
or simply rejected it once they were within its hydrodynamic influence (Weber et al.,
2006). Poor performance of bypass collectors designs is most commonly the result of the
uncertainty about the necessary flow characteristics to attract migrants to the vicinity of
the collector and entice them to enter (Goodwin, 2003).
Schilt (2006) defines the routes of fish passage through hydropower dams as turbine passage, juvenile bypass system passage, spillway passage and surface passage.
Passage of migrants through turbines is generally considered the passage most likely to
injure or kill fish. Not all fish are injured or killed as they pass through turbines, the
mortality rate or injury can vary with time of year, species, age, water temperature, turbine
type and other factors. Mechanisms which can injure or kill migrants include pressure
changes, shearing between different moving water masses, impact with turbine or dam
structures, cavitation, grinding and abrasions. A great deal has gone into making turbines
more fish friendly (Schilt, 2006).
Most bypass systems for juvenile salmon at major hydroelectric projects involve screening juveniles from deep turbine intakes (Coutant & Whitney, 2000). The screens located
in the turbine intakes of dams, usually rotating conveyor belt like mesh screens, are designed to deflect flow and fish to the upper portion of the turbine intake and upward into a
passage called a gatewell slot, a vertical passageway. From the gatewell slot the fish is
then passed through an orifice into a juvenile bypass system, were the actual bypass is
often a modification of ice and sluiceways (Schilt, 2006).
Passage of migrants through spillways is considered the most benign passage route for juvenile salmon. Disadvantages of spill include elevated dissolved gas downstream which
can stress or kill fish as well as reducing swimming performance and resistance to pathogens. Spillway operations also have the drawback of large amount of spill that is foregone
power generation (Schilt, 2006).
Surface passage routes or surface flow outlets (SFO) are designed to collect fish near the
surface of near-dam forebay and pass the fish into the tailrace. The design is considered
cost effective in terms of fish passage and discharge. In the Columbia and Snake river
systems too few fish use these routes, relatively small portion of the total downstream
migration passes. If the system and design work properly this is thought to be the most
benign option of downstream migrant passage system (Schilt, 2006).
Since 1981 the USACE has been collecting juvenile salmon from bypass systems of several dams. The bypass is partially dewatered and rather elaborate system of gates enable the
10
11
1 Introduction
Wells Dams started operation in 1967, the power plant is a run of the river plant with
mean discharge of 6230 m3 /s and installed power generating capacity of 840 MW. The
SFO at Wells Dam was developed in the years 1983 to 1990 and is one of the most
successful SFO located on the Columbia river. Research at Wells Dam has showed that
up to 89% of juvenile salmon goes into the SFO and their mean survival rate is about 96%
(Sweeney et al., 2007).
The Cowlitz Falls powerplant started operation in 1994. It has a mean discharge of 150
m3 /s and installed power generating capacity of 70 MW. The SFO at Cowlitz Falls was
built at the same time as the dam and was designed with the successful SFO at Wells Dam
in mind. The SFO at Cowlitz Falls has not proved as successful as the SFO at Wells Dam
which is considered to be the cause of irregularities in the flow field in front of the SFO
entrance. Research has shown that only 48% of juvenile steelhead salmon has entered the
SFO at Cowlitz Falls and the success for other species is less. No research is available on
the survival rate of juvenile fish at Cowlitz Falls. In recent years considerable effort has
been put into making the SFO at Cowlitz Falls more successful (Sweeney et al., 2007).
12
13
2 Methods
2.1 Theory
The theory and equations presented in this section are derived from the ANSYS Solver
Theory manual (ANSYS, 2011).
+ ( U ) = 0
(2.1)
t
where is density and U is velocity magnitude. The momentum equations are defined as
( U )
+ ( U U ) = p + + SM
(2.2)
t
where p is static pressure, SM a momentum source and the stress tensor which is related
to the strain rate by
2
T
= U + (U ) U
(2.3)
3
where is dynamic viscosity, T static temperature and the Kronecker delta function.
The last of the governing equations the total energy equation is expressed as
( htot ) p
(2.4)
+ ( U htot ) = ( T ) + (U ) +U SM + SE
t
t
where is thermal conductivity, SE is an external energy source and htot is the total
enthalpy. The term (U ) represents the work due to viscous stresses and is called the
viscous work term. The term U SM represents work due to external momentum sources
and is currently neglected. The total enthalpy htot is related to the static enthalpy h(T, p)
by
1
(2.5)
htot = h + U 2
2
For further details of ANSYS CFX solver theory see (ANSYS, 2011).
15
2 Methods
2.1.2 Turbulence Models
Turbulence can be described as fluctuations in a flow field in time and space. The process
is very complex and occurs when inertia forces exceed viscous forces in a fluid at high
Reynolds numbers. Turbulence length and time scales at realistic Reynolds numbers span
a large range, the smallest much smaller than the finest finite volume mesh. Though
the Navier-Stokes equations describe both laminar and turbulent flow without additional
information, the process of solving the equations directly is impractical and would require
computing power of an order of magnitude higher than is available today. Thus great
amount of research in numerical modelling has gone into predicting turbulence with the
use of turbulence models. The turbulence models have been developed to account for the
effects of turbulence without extremely fine mesh and direct numerical simulation and are
in most cases statistical models (ANSYS, 2011).
The turbulence model used in this study is the k- turbulence model of CFX, for further
details see (ANSYS, 2011).
= 1 NP
16
(2.6)
2.1 Theory
where is a general scalar variable, indicates phase and NP is the total number of
phases. Because transported quantities are shared the shared fields can be solved using
bulk transport equations instead of solving individual phasic transport equations. The bulk
transport equations are derived by summing the individual phasic transport equations over
all phases into a single transport equation for
( ) + ( U ) = S
(2.7)
t
where is density, U velocity, is diffusivity defined and expressed in terms of volume
fractions r of phase as
NP
(2.8)
1
r U
=1
(2.9)
=1
NP
U=
NP
(2.10)
=1
(2.11)
( U ) + U U U + (U )T = SM p
(2.12)
(2.13)
=1
The continuity equation with volume fractions taken into account becomes
NP
(r ) + (r U ) = SMS +
t
=1
(2.14)
where SMS describes user specified mass sources and the mass flow rate per unit
volume from phase to phase . This term only occurs if interface mass transfer takes
place. The homogeneous volume conservation equation is simply the constrain that the
volume fractions sum to unity
NP
r = 1
(2.15)
=1
Combining equation with the continuity equation yields the volume continuity equation
solved by the CFX-Solver
!
NP
1
1
(2.16)
t (r ) + (r U ) = SMS +
=1
For further details see (ANSYS, 2011).
17
2 Methods
2.1.4 Numerical Solver
The coupled solver of ANSYS CFX was used which solves the hydrodynamic equations
(for u, v, w, p) as a single system. The coupled solver uses a fully implicit discretization
of the equations at any given time step which for steady state problems the time step
acts as an acceleration parameter to guide the approximate solutions in a physically based
manner to a steady state solution. This approach reduces the number of iterations required
for convergence to a steady state (ANSYS, 2011).
18
19
2 Methods
20
21
2 Methods
Conveyance channel should be as smooth as possible, no intrusions, sharp
corners or rough walls
Ensure sediment or debris can not accumulate in channel
No deceleration or local acceleration more than 0.3 m/s per meter
No hydraulic jump inside channel
Pressure larger than or equal to atmospheric pressure
Outfall
Locate were flow velocity in river is more than 1.2 m/s
No vortices or reverse flow
Sufficient depth eliminating potential collision of juvenile salmon with riverbed
Maximum velocity of 7.6 m/s
22
Figure 2.2: Overview of 3D model of Heiarln reservoir and structures used in numerical model computation.
23
2 Methods
The final stage in preparation for meshing is to extract the fluid volume from the geometry. A second solid is made which covers the whole domain and the topography solid
subtracted from the second solid, see Figures 2.4 and 2.5.
Figure 2.4: Volume of fluid subtracted from geometry of reservoir used as input for meshing process.
Figure 2.5: Volume of fluid subtracted from SFO model geometry used as input for meshing process.
24
25
2 Methods
26
water = 1 air
(2.18)
The physical properties of the numerical domain are then defined, the two fluids air and
water are predefined with normal properties, for the water phase a buoyancy model is
activated and a turbulence model selected. The two models in this study the reservoir
and the SFO models have similar boundary conditions. The reservoir model has extra
outlet boundaries for the power intakes and the spillway gates. The inlet boundaries are
defined as a velocity inlets, with normal speed divided uniformly over the boundary. For
the inlet boundary the volume fractions of both fluids are also defined and the turbulence
option set as zero gradient. For the SFO outlets a static pressure boundary is defined. The
outlets for the power intakes and spillways are defined as mass flow outlets where the
total mass flow to the turbines is divided evenly between the four outlets. The top of the
numerical domain is defined as an opening boundary where the volume fraction of air is
always one but water always zero. The topography and walls of the structures are defined
as noslip walls, smooth with no roughness.
When the boundary conditions have been defined initial conditions and the solver set up
is defined. As an initial condition a pressure condition is defined corresponding to NRWL
in all the numerical domain. Initial flow velocities were defined as automatic.
For the solver control a first order advection scheme was used for both models. For Cases
B and C of the reservoir model a specified blend factor was specified for the advection
scheme, a low value 0.1 was used for increased robustness. For Case A of the reservoir
model and all cases of the SFO model a high resolution advection scheme was used. A
local timescale factor for timescale control was used for Cases B and C of the reservoir
model while a physical time scale factor was used for Case A of the reservoir model
and for all cases of the SFO model. The local timescale factor option enables the use
27
2 Methods
of different time scales at different regions in the computational domain. The local time
scale factor which is entered is a multiplier of a local element based time scale where
smaller time scales are applied to regions of fast flow and larger time scales at slower
flow regions. A local timescale factor of 0.3 was used. With the physical timescale, a
fixed time scale is used for the entire computational domain. The physical timescale was
used to provide sufficient relaxation of the equation nonlinearities to obtain a steady state
solution. A physical timescale of 0.3 s was used for Case A in the reservoir model while
a physical timescale of 0.1 s was used for the SFO model (ANSYS, 2011). To reduce
the simulation time of Case B in the reservoir model, the results of Case C were used as
initial conditions for the Case B simulation.
2.4.4 Execution
The numerical models were calculated with double precision, the models were iterated
until a specific convergence criteria had been reached which differed between the reservoir
and SFO models.
The reservoir model was iterated until the residuals for the root mean square for the governing equations had reached a value of 1104. The stability of velocity inside the
approach flow channel was also used as a convergence indicator. For the SFO model the
convergence criterion was specified as 1% global mass imbalance for the water phase and
steady massflow of water at the outlet boundary.
The size and complexity of the reservoir model topography resulted in much longer
computational time of the numerical simulation. In Table 2.1 the total iterations and
accumulated time of simulation are shown. The simulation times are quite variable for
the reservoir cases, Case A took almost two days longer to simulate than Case C because
of the different advection scheme and timescale control where a more robust approach
was used for Cases B and C. For reservoir Case B, Case C results were used as initial
conditions for the simulation which reduced the simulation time down to approximately
2 days. For the SFO cases small difference in simulation time is observed which is due
to different computers used in the calculations. The SFO cases labelled Q20 to Q50 refer
to the discharge series tested, that is Q20 refers to the case where the SFO discharge is
20 m3 /s, Q30 to a discharge of 30 m3 /s and so forth. Cases Q20 and Q30 were computed using a laptop equipped with a 2.8 GHz Intel i7 processor while Cases Q40 and Q50
where computed using a desktop equipped with 3.4 GHz Intel i7 processor. The laptop
had problems with cpu cooling which reduced the performance of the processor while
the desktop cpu maintained steady performance because of better cooling. All reservoir
simulations where conducted using the aforementioned desktop.
28
Figure 2.11: The intake and SFO structure of the physical model.
Discharge in the model was regulated with two pumps with frequency inverters. Two
reservoir tanks, one downstream and one upstream, collected the water which was circulated in a closed loop as shown in Figure 2.13. High accuracy ultrasonic sensors measured the discharge in the system and manual gauges measured pond water elevations and
monitored stability. A flow straightness structure was located at the outlet of the upstream
reservoir tank. The flow straightness structure was used to direct the flow entering the
approach flow channel more along the right approach bank which was in accordance with
29
2 Methods
preliminary results from the numerical model.
Measurement methods used to describe and quantify the behaviour of the system consisted of particle tests, dye tests, velocity measurements and visual observations documented with photographs and videos.
The physical model study and measurements techniques are described in more detail in
(Tomasson et al., 2012).
30
31
2 Methods
2.5.1 Velocity Measurements
Velocity measurements were conducted using a Sontek ADV which measures the three
velocity components. Each measurement lasted for 60 seconds, collecting 300 data points. The ADV, which is downward looking, has a blind zone of 50 mm from the sensors
probe (Sontek, 1997). Because of the blind zone measurements were limited to a depth
exceeding 2 m (prototype depth) from the surface. ADV measurements were made at
a depth of 2.6 m at points divided on a 10 m x10 m rectangular grid closest to the
SFO entrance and 20 m x 20 m grid further away from the SFO, a total of 51 points. ADV data processing and filtering was done with USBRs WinADV32 software
(Bureau of Reclamation & Group, 2011).
32
33
2 Methods
34
35
Figure 3.1: Overview of the Urriafoss HEP original design layout (Landsvirkjun, 2010).
Figure 3.2: A longitudinal view of the original SFO design (Landsvirkjun, 2010).
36
Figure 3.3: A plan view of the original SFO design (Landsvirkjun, 2010).
Figure 3.4: Revised layout of the approach channel with modifications from February
2012 (Landsvirkjun, 2010).
37
Figure 3.5: A longitudinal view of the revised SFO design (Landsvirkjun, 2010).
Figure 3.6: A plan view of the revised SFO design (Landsvirkjun, 2010).
38
Figure 3.8: Contour and vector plot from the preliminary numerical model showing flow diversion from left approach bank.
In light of the irregularities forming at the left approach bank modifications were made
in attempt to get the flow more along the left approach bank limiting the formation of
vortices and stagnant velocity zones along the left approach bank. The modifications
consisted of changing the left approach bank in order to get the water flowing from the
left approach bank to enter the approach channel parallel to the main current. The angle of
the left abutment of the power intake was also increased to limit the potential of a stagnant
velocity zone forming at the left abutment.
39
40
Figure 3.9: Overview of the Urriafoss HEP final design: (1) the original riverbed, (2)
spillway approach flow channel, (3) intake and SFO approach flow channel, (4) intake to
the power house and SFO structure, (5) the spillway structure, (6) the fuse plug, (7) the
upstream fishway, (8) the mandatory release structure, (9) Urriafoss Dam (Landsvirkjun,
2010)
41
Figure 3.10: Top view and longitudinal section of the intake and SFO structure (Landsvirkjun, 2010)
42
3.2 Reservoir
3.2 Reservoir
The reservoir model was tested with respect to three operational scenarios, Cases A, B and
C corresponding to Cases 1.2, 1.3 and 1.4 set forth by Karadttir and Gudjonsson (2012b).
The cases represent normal operational conditions expected at the time of migration of
juvenile salmon. Measurements showed the SFO discharge capacity being about 30%
lower than the calculated design discharge at NRWL, 50 m a.s.l. Scale effects are only
considered to account for small amount of the reduction in discharge capacity as discussed in Section 2.5.5. Some unconventional features of the SFO, the inward angle of the
structure and location of power intake below the SFO crest, are not incorporated into the
conventional discharge capacity equation for ungated spillways which probably accounts
for most of the difference in measured and calculated discharge capacity. This is not
explored further in the thesis but should be considered in the final design of the structure.
To ensure similar conditions in the numerical model a reservoir elevation of 50.2 m a.s.l.
was used. Discharges for each case and the percentage of time of equal or more discharge
are presented in Table 3.1. In Figure 3.11 the flow duration curve derived from a discharge
series spanning 55 years, 1950 to 2005, is shown. The flow duration curve is derived for
the time period which the downstream migration of juvenile salmon occurs in the Lower
jrs River, which is annually from the 15th of May to the 15th of June. As shown in
the figure 53% of the time the discharge is equal or more than the discharge going to the
power intake (370 m3 /s) and the environmental flow (10 m3 /s) combined. This means
that in 53% of the time there should be sufficient amount of flow entering the Heiarln
Reservoir to supply water to the SFO. According to the flow duration curve no discharge
is routed through the spillway in 59% of the time. Karadttir and Gudjonsson (2012a)
found out that 88% of the juvenile salmon entering the reservoir would be attracted towards the SFO and 12% would be attracted towards the spillway. They assumed the juvenile
salmon run would divide between the structures in direct relation to the discharge routed
through each structure. They also assumed the magnitude of the salmon run each day to
be directly related to the discharge, that is more juveniles migrating in more flow. When
they assumed the juvenile salmon to migrate evenly over the migration period 91% of the
fish would enter the SFO and 9% would be attracted towards the spillway.
Table 3.1: Overview of the cases tested in the physical and the numerical models.
RWL
QIntake QSpillway QSFO QTotal
Percentage of time of
Case
[m a.s.l.] [m3 /s]
[m3 /s] [m3 /s] [m3 /s] equal or more discharge
A
50.2
370
0
40
410
41 %
B
50.2
370
70
40
480
25 %
C
50.2
370
235
40
645
5%
43
1400
Discharge [m /s]
1200
1000
800
600
400
200
0
0
10
20
30
40
50
60
70
80
90
100
Figure 3.11: Flow duration curve for time period where annual downstream migration
occurs, 15th of May to 15th of June. The purple line shows percentage of time when the
discharge is equal or more than the intake and environmental flow combined while the
green line shows the time when the discharge is greater than the intake, the SFO and
environmental flow combined.
Figure 3.12: Section lines of measurements in physical model used in validation of numerical model.
44
3.2 Reservoir
3.2.1 Case A
In Figure 3.13 velocities from ADV measurements in the physical model (dots) and results
from the numerical model (solid line) taken at 2.6 m depth, are compared at Section lines
10, 20, 40, 60, 80 and 100 for Case A; RWL = 50.2 m a.s.l., QIntake = 370 m3 /s and QSFO
40 m3 /s. The ADV measurement points are shown with one standard deviation error
band to indicate the magnitude of the flow fluctuations.
At Section 10, shown in Figure 3.13, a slow moving water is observed in front of the
spillway entrances with gradually increasing velocity from Station 70 to a maximum
velocity in front of the intake. The physical and numerical data differ most between
Stations 80 and 90 where two peaks are observed in the numerical model right in front of
the curb between the spillway and intake but are otherwise in good agreement.
As at Section 10, low velocities are observed in front of the spillway at Section 20 as seen
in Figure 3.13. The velocity at Section 20 gradually increases from Station 50 towards the
intake structure reaching a maximum for the physical model at Stations 110 and 120 from
where the velocity decreases again at Station 130. For the numerical model the velocity
takes an absolute maximum value at Station 130 opposite to the decrease in velocity
observed in the physical model.
At Section 40, the ADV measurements from the physical model show that the velocity
gradually increases from Station 30 to Station 90 where maximum value is reached. From
Station 90 to 110 the velocity is similar but from Station 110 a gradual decrease is observed in the physical model. The character in the numerical model at Section 40 is in good
agreement with the physical model results from Station 30 to Station 110 where the results
start to differ. Note the higher standard deviation of the ADV measurements at Stations
110 to 130 which is due to irregularities forming at the left approach bank (looking downstream) affecting the measurements. The irregularities formed where discharge flowing
over the left approach bank enters the approach flow channel perpendicular to the main
current in the approach flow channel. Small topographic features upstream of the left
approach bank also contribute to the formation of irregularities.
For the ADV measurements from the physical model at Section 60 the velocity gradually
increases from Station 30 to Station 90 where a maximum value is reached. From Station 90 the velocity gradually decreases again to 0.28 m/s at Station 130. The numerical
model has a rather different character at Section 60. Instead of a U-shape distribution, the
velocity gradually increases to an absolute maximum at Station 122 which is somewhat
higher than observed in the physical model. The standard deviation of the ADV measurements is higher for the values at Stations 110 to 130 as in Section 40 due to the same
artefact as discussed for Section 40.
At Section 80, the difference between ADV measurements and numerical model results
45
46
3.2 Reservoir
flow streamlines (three dimensional streamlines changing both laterally and vertically)
while the plot on the right shows surface streamlines. The bulk flow and surface streamlines represent consistent flow behaviour over the depth of the numerical model. The
models are in good agreement, in both figures the extent of the attraction flow is great as
most of the approach channel flow is drawn towards the intake and the SFO. Two zones of
irregularities are observed in the physical model, labelled stagnant zone and vortex zone
in Figure 3.16. The stagnant zone located immediately upstream of spillway is occupied
by a slow moving water body. Particles entering the zone may linger for some time until
finally drawn towards the intake and the SFO. In the vortex zone located by the left approach bank irregularities form where different currents intersect with steady formation of
small shallow vortices.
47
Section 10
Section 20
0.8
Velocity Magnitude [m/s]
0.8
0.6
0.4
0.2
0
20
0.6
0.4
0.2
0
40
60
80
100
Station [m]
120
140
20
40
Section 40
0.4
0.2
120
140
120
140
0.6
0.4
0.2
0
40
60
80
100
Station [m]
120
140
20
40
Section 80
60
80
100
Station [m]
Section 100
0.8
Velocity Magnitude [m/s]
0.8
Velocity Magnitude [m/s]
140
0.8
0.6
0.6
0.4
0.2
0
20
120
Section 60
0.8
20
60
80
100
Station [m]
0.6
0.4
0.2
0
40
60
80
100
Station [m]
120
140
Numerical Model
20
40
60
80
100
Station [m]
Physical Model
Figure 3.13: Velocity comparison between physical and numerical models taken at 2.6 m
depth at Section Lines 10, 20, 40, 60, 80 and 100 for Case A: RWL = 50.2 m a.s.l., QIntake
= 370 m3 /s and QSFO 40 m3 /s.
48
3.2 Reservoir
Figure 3.14: Velocity contours and vectors at 2.6 m depth from the physical model for
Case A: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s and QSFO 40 m3 /s.
Figure 3.15: Velocity contours and vectors at 2.6 m depth from the numerical model for
Case A: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s and QSFO 40 m3 /s.
49
Figure 3.16: Results from particle test for Case A, QIntake = 370 m3 /s, QSFO = 40 m3 /s
and QSpillway = 0 m3 /s. Lines with arrows represent general flow characteristics. Colours
refer to zones described in section 2.2, the decision zone is shown in orange and the
discovery zone in yellow.
Figure 3.17: Streamlines in approach flow channel from the numerical model, Case A.
On left: Bulk flow streamlines. On right: Surface streamlines.
50
3.2 Reservoir
In Figure 3.18 velocity magnitude contours and vectors are shown at an elevation of 49.5
m a.s.l., 0.7 m depth, in the numerical model. As the physical and numerical model are
in good agreement for Case A, as seen in Figures 3.14 and 3.15 for water depth 2.6 m,
it can be assumed that the numerical model represents accurately the conditions at this
shallower depth, 0.7 m, a depth where the juvenile salmon is likely to inhabit. As seen
in Figure 3.18 the conditions at 0.7 m depth are very similar to the conditions at 2.6 m
depth in Figure 3.15, the extent of the attraction flow is extensive with gradual increase
in velocity towards the intake and the SFO. The only difference observed is the stagnant
velocity in front of the spillway extends somewhat further upstream than observed at 2.6
m depth.
Figure 3.18: Velocity contours and vectors at 0.7 m depth from the numerical model for
Case A: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s and QSFO 40 m3 /s.
51
52
3.2 Reservoir
At Section 100, the maximum values of the numerical model are reached at Station 50, for
the physical model a maximum value is reached at Station 130. At these two stations the
models differ most. All numerical values are within one standard deviation of the ADV
measurements with the exception of Station 50. The models are thus in good agreement
at Section 100. The most fluctuation is observed at Station 130 as before as the standard
deviation shows.
In Figure 3.19 the characteristic flow behaviour is similar at all section lines and the
models are in good agreement especially for Sections 10 and 20. For Sections 40 to 100
the character is similar with somewhat lower velocities observed in the physical model.
At Section 40 a high value of standard deviation is observed at Station 120, for Sections
60 to 100 high standard deviation values are observed at Station 130 which are caused by
irregularities forming at the left approach bank.
Figures 3.20 and 3.21 show velocity contours and vectors for the physical and numerical
models, respectively, at 2.6 m depth. The models show similar characteristic flow behaviour, the stagnant velocity zone in front of the spillway has reduced considerably in both
models from Case A. The stagnant zone extends upstream along the right approach bank.
Maximum values are observed in front of the intake and the SFO in both models. In the
numerical model high velocities are observed at right approach bank where water falls
over the bank. The main current is located in the center of the approach flow channel, the
current reaches further downstream than in Case A as is expected with the added spillway
discharge. The main current is headed towards the intake and the SFO as in Case A. Irregularities are observed at the left approach bank as in Case A, the same characteristic
difference between models is also observed as in Case A. Although visually the physical
and the numerical model are in a better agreement for Case B than Case A. The figures
show extensive attraction flow towards the SFO and the intake with gradual increase in
velocity towards the structures. In both models the extent of the stagnant velocity zone
in front of the spillway has decreased considerably from Case A due to of the spillway
discharge. In the figures, the extensive attraction flow and gradual increase in velocity
towards the intake and the SFO is clear which should be ideal for attracting juvenile
salmon towards the SFO.
The vertical extent of the stagnant velocity zone is shown in Figure 3.22 where velocity
contour lines and velocity vectors are shown at a longitudinal section perpendicular to
spillway gate 2. As the figure shows the stagnant zone extends to depths of approximately
4 m immediately upstream of the spillway crest and the extent laterally is also considerable. Fish attracted towards the spillway may loose meander for some time in front of the
spillway gates as they generally dive through deep passages as last resort. The distance
from the stagnant zone to the point where the surface flow starts to accelerate towards the
SFO is not great. Assuming that juvenile salmon is active in search of appropriate route
through the system they should be able to find the SFO attraction flow.
53
Section 10
Section 20
0.7
Velocity Magnitude [m/s]
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.1
20
0.6
0.5
0.4
0.3
0.2
0.1
0
40
60
80
100
Station [m]
120
0.1
20
140
40
Section 40
0.5
0.4
0.3
0.2
0.1
0
120
140
120
140
0.6
0.5
0.4
0.3
0.2
0.1
0
40
60
80
100
Station [m]
120
0.1
20
140
40
Section 80
60
80
100
Station [m]
Section 100
0.7
Velocity Magnitude [m/s]
0.7
Velocity Magnitude [m/s]
140
0.7
0.6
0.6
0.5
0.4
0.3
0.2
0.1
0
0.1
20
120
Section 60
0.7
0.1
20
60
80
100
Station [m]
0.6
0.5
0.4
0.3
0.2
0.1
0
40
60
80
100
Station [m]
120
140
Numerical Model
0.1
20
40
60
80
100
Station [m]
Physical Model
Figure 3.19: Velocity comparison between physical and numerical models taken at 2.6 m
depth at Section Lines 10, 20, 40, 60, 80 and 100 for Case B: RWL = 50.2 m a.s.l., QIntake
= 370 m3 /s, QSFO 40 m3 /s and QSpillway = 70 m3 /s.
54
3.2 Reservoir
Figure 3.20: Velocity contours and vectors at 2.6 m depth from the physical model for
Case B: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s, QSFO 40 m3 /s and QSpillway = 70
m3 /s.
Figure 3.21: Velocity contours and vectors at 2.6 m depth from the numerical model for
Case B: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s, QSFO 40 m3 /s and QSpillway = 70
m3 /s.
55
Figure 3.23: Velocity contours and vectors at 0.7 m depth from the numerical model for
Case B: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s, QSFO 40 m3 /s and QSpillway = 70
m3 /s.
56
3.2 Reservoir
Figures 3.24 and 3.25 show general flow behaviour and streamlines in the approach flow
channel for the physical and numerical models respectively, for Case B. In Figure 3.25
bulk flow stream lines are shown on the left while surface streamlines are shown on the
right. The models are in good agreement as the separation of streamlines heading towards the spillway is very similar between models, especially the particle tracks from the
physical model and the bulk flow streamlines. A circular movement is observed close
to the right approach bank in Figure 3.25. Although the movement is not presented in
Figure 3.24, weak circular movement was documented during the particle test in the
physical model just upstream from the location where the circular movement is observed in the numerical model. From Figures 3.24 and 3.25 the separation of streamlines
is clear indicating the intake and the SFO as dominating features influencing the overall
approach flow character.
Figure 3.24: Results from particle test for Case B., QIntake = 370 m3 /s, QSFO = 40 m3 /s
and QSpillway = 70 m3 /s. Lines with arrows represent general flow characteristics, the
decision zone is shown orange and the discovery zone yellow.
57
Figure 3.25: Streamlines in approach flow channel from the numerical model, Case B.
On left: Bulk flow streamlines. On right: Surface streamlines.
3.2.3 Case C
In Figure 3.26 velocities from ADV measurements in the physical model (dots) and results
from the numerical model (solid line) taken at 2.6 m depth are compared at Section Lines
10, 20, 40, 60, 80 and 100 for Case C: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s, QSFO
40 m3 /s and QSpillway = 235 m3 /s. The ADV measurement points are shown with one
standard deviation error band to indicate the magnitude of the flow fluctuations.
At Section 10 shown in Figure 3.26 the velocity distribution in the physical model is consistent with the numerical model results. A relatively even distribution is seen between
Stations 30 to 70 with the exception of Station 30 in the physical model where a considerably lower value is observed. At Station 80 the velocity decreases in both models but
from thereon increase to a maximum value at Station 110.
At Section 20 the velocity distribution is similar to the distribution at Section 10, the
velocity from Station 30 to 80 has increased a bit for the numerical model. The models are
in good agreement for Stations 50 to 120. At Stations 30, 40 and 130 the physical model
show considerably lower values than the numerical results though the numerical values
at Station 130 are within one standard deviation of the physical model value. Maximum
value is reached for both models at Station 120.
At Section 40 the velocity distribution differs between models with lower values in most
cases observed in the physical model. The physical model velocity distribution is in a
U-shape manner whereas the numerical velocity distribution takes minimum values at
each side of the approach flow channel and has a flat top from Station 40 to Station 120.
Numerical values are only within one standard deviation of the physical model values at
Station 70 and Station 110.
At Section 60 a U-shape velocity distribution is observed in both models with the nu-
58
3.2 Reservoir
merical model having a flatter top. The models differ in values at Stations 30, 110 and
130 where the physical model takes considerably lower values. At other locations numerical results are within one standard deviation of the physical model results.
At Section 80 the models are in good agreement with the exception of the physical model
value at Station 30 being considerably lower. The lower value at Station 30 has high
standard deviation indicating fluctuations caused by turbulence in the model. The velocity
distribution is relatively flat over the whole section with values ranging between 0.4 m/s
and 0.6 m/s.
At Section 100 the models are consistent at Stations 50 and 70, at Stations 90 and 110
the velocity is considerably lower in the physical model. At Station 130 the ADV measurements take a maximum value of 0.8 m/s which is higher than the numerical model
value at same location. The overall behaviour at Section 100 is consistent between the
models if the outlier at Station 130 is ignored.
In Figure 3.26 the character between the models is very similar for Sections 10, 20 and
80. For Sections 40 and 60 the character differs more between the models with physical
model velocity distribution in a somewhat U-shape manner while the numerical represents
a more even velocity distribution. At Section 100 at noticeable outlier is observed at
Station 130 at the left approach bank.
Figures 3.27 and 3.28 show velocity contours and vectors in the approach flow channel
in the physical and numerical model respectively at 2.6 m depth. A similar behaviour is
observed in the approach flow channel in both models, the stagnant velocity zone in front
of the spillway has almost diminished and the approach flow spreads over the width of the
approach flow channel. The approach flow channel is split into two parts, one is occupied
by the current heading towards the intake and the SFO, the other heading towards the
spillway. The main difference between the models is as in Cases A and B a strong current
represented by a large velocity component at the left approach bank in the physical model
perpendicular to the main current in the approach flow channel. Small stagnant velocity
zones are observed inside the spillway bays in the numerical model as the water is pulled
down under the spillway gates. In the figures the extensive attraction flow and gradual
increase in velocity towards the intake and the SFO is clear. In both models the extent
of the stagnant velocity zone in front of the spillway has reduced and almost disappeared
because of the increased spillway discharge. The whole approach flow channel is now
occupied by the main approach flow current.
The vertical extent of the stagnant velocity zone is shown in Figure 3.29 where velocity
contour lines and velocity vectors are shown at a longitudinal section perpendicular to
spillway gate 2. As the figure shows the stagnant zone extends to depths of approximately
2 m immediately upstream of the spillway crest, the lateral extent of the stagnant zone
has decreased from Case B and occupies almost exclusively the area within the spillway
bays. Fish attracted towards the spillway may loose meander for some time in front of the
59
60
3.2 Reservoir
Section 10
Section 20
1
Velocity Magnitude [m/s]
1
0.8
0.6
0.4
0.2
0
20
40
60
80
100
Station [m]
120
0.8
0.6
0.4
0.2
0
20
140
40
Section 40
0.6
0.4
0.2
40
60
80
100
Station [m]
120
120
140
120
140
0.8
0.6
0.4
0.2
0
20
140
40
Section 80
60
80
100
Station [m]
Section 100
1
Velocity Magnitude [m/s]
1
Velocity Magnitude [m/s]
140
0.8
0.8
0.6
0.4
0.2
0
20
120
Section 60
0
20
60
80
100
Station [m]
40
60
80
100
Station [m]
120
140
Numerical Model
0.8
0.6
0.4
0.2
0
20
40
60
80
100
Station [m]
Physical Model
Figure 3.26: Velocity comparison between physical and numerical models at Section
Lines 10, 20, 40, 60, 80 and 100 for Case C: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s,
QSFO 40 m3 /s and QSpillway = 235 m3 /s.
61
Figure 3.27: Velocity contours and vectors at 2.6 m depth from the physical model for
Case C: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s, QSFO 40 m3 /s and QSpillway = 235
m3 /s.
Figure 3.28: Velocity contours and vectors at 2.6 m depth from the numerical model for
Case C: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s, QSFO 40 m3 /s and QSpillway = 235
m3 /s.
62
3.2 Reservoir
Figure 3.30: Velocity contours and vectors at 0.7 m depth from the numerical model for
Case C: RWL = 50.2 m a.s.l., QIntake = 370 m3 /s, QSFO 40 m3 /s and QSpillway = 235
m3 /s.
63
Figure 3.31: Results from particle test for Case C, QIntake = 370 m3 /s, QSFO = 40 m3 /s
and QSpillway = 235 m3 /s. Lines with arrows represent general flow characteristics, the
decision zone is shown orange and the discovery zone yellow.
Figure 3.32: Streamlines in approach flow channel from the numerical model, Case C.
On left: Bulk flow streamlines. On right: Surface streamlines.
64
3.2 Reservoir
3.2.4 Vertical Streamline Separation between the SFO and the Intake
In Table 3.2 results of the dye tests conducted during the physical model test are shown.
The result show the depth of water which the SFO transports ranges between 1 m and 1.5
m for cases A, B and C. The depth at which half of the water is transported by the SFO
and half by the intake ranges between 1.5 m and 2 m. The depth at which the intake starts
solely to draw water is below 2 m depth.
Table 3.2: Results from dye tests showing Reservoir Water Level (RWL), depth where
water flows only to the SFO (Only SFO), depth where water is divided evenly between the
intake and the SFO (Intake/SFO; 50/50) and the depth where water flows only to intake
(Only Intake).
Case
A
B
C
RWL
[m a.s.l.]
50.2
50.2
50.2
Only SFO
[m] depth
1-1.5
1-1.5
<1
Intake/SFO; 50/50
[m] depth
1.5-2
1.5-2
1.5
Only Intake
[m] depth
>2
>2
>2
During the dye test a distinct behaviour was observed immediately upstream of the SFO
crest where a dye released perpendicular to SFO Entrances 1 and 4 was drawn toward the
center entrances, Entrances 2 and 3. The observed behaviour shown in Figure 3.33 may
be caused by the inward angle of the intake and SFO structure.
Figure 3.33: Streamlines drawn towards SFO Entrances 2 and 3 during a dye test.
In Figure 3.34 streamline separation in front of THE intake and the SFO is shown. In the
figure two plots are shown, on left a plot extending the whole depth in front of the intake
and the SFO and reaching 25 m upstream, on right a close up of the streamline separation
65
51
50
50.5
48
50
49.5
Elevation [m a.s.l.]
Elevation [m a.s.l.]
46
44
42
40
38
49
48.5
48
47.5
36
34
47
32
46.5
30
10
15
20
X [m]
25
46
1
X [m]
Figure 3.34: Streamline separation between the intake and the SFO in the numerical
model for Case A. The blue streamlines head to the intake and the red streamlines to the
SFO. On the left, the streamline separation over the whole depth in front of the intake is
shown and on the right a more detailed view of the streamline separation at the SFO nose
is shown. The detailed area is shown on the left as a box with dashed lines.
66
3.2 Reservoir
2.3, smooth acceleration about 1 m/s per meter just upstream of the SFO entrance and
flow velocities at the SFO entrance > 2-3 m/s.
3
SFO Entrance 1
SFO Entrance 2
SFO Entrance 3
SFO Entrance 4
2.5
2
1.5
1
0.5
0
4
6
Distance from SFO entrance [m]
10
Figure 3.35: Velocity magnitudes in the Decision zone at lines perpendicular to the SFO
entrances for Case A.
3.2.6 Summary
The models represent a prediction of conditions in the discovery and decision zones which
can be used to evaluate the effectiveness of the approach and the SFO layout. In general,
the approach flow conditions for all cases are satisfactory and the models are in good
agreement regarding the general flow behaviour. The approach flow conditions inside the
approach flow channel are influenced by project operations, when spillway discharge is
zero the intake and the SFO are the main factors influencing the approach flow conditions.
This good performance of the SFO can be expected in 59 % of the time as shown in
Table 3.1. With increasing spillway discharge the spillway influence increases. It can
therefore be concluded that the approach flow conditions are directly related to the ratio
between the spillway and the intake discharge. In the stagnant velocity zone in front of the
spillway juvenile salmon might get delayed, losing track of the main current towards the
SFO. Likewise during spillway operation fish attracted towards the spillway may enter
the stagnant zone in front of the spillway and loose track of the SFO attraction flow.
However, this is in most cases a small portion of the flow and events of significant amount
of spillway discharge as in Case C can only be expected over 5 % of time as shown in
Table 3.1.
The physical and numerical models are consistent with regards to streamline separation in the approach flow channel. The bulk flow streamlines from the physical model
67
68
Figure 3.36: SFO channel naming convention and cross section location.
69
Elevation [m a.s.l.]
53
52
51
50
49
48
47
46
45
0
10
15
20
25
30
35
40
45
50
55
60
Station [m]
3
Q = 20 m /s
Q = 30 m /s
Q = 40 m /s
Q = 50 m /s
Figure 3.37: Water levels inside the SFO channel at a cross section shown in Figure 3.36.
In Figure 3.40 water levels inside the SFO at longitudinal sections perpendicular to each
SFO entrance are shown. The water levels range from 47.2 m a.s.l. to 49.4 m a.s.l., which
corresponds to 2.2 m to 4.4 m water depths, respectively. The water levels downstream of
the SFO crest rise above the SFO crest elevation, 49.1 m a.s.l., for discharges, 40 m3 /s and
50 m3 /s at Entrances 3 and 4. At 50 m3 /s discharge water levels downstream of entrance 2
rise also above the SFO crest elevation. Though the water elevations rise high during 40 to
50 m3 /s discharge the downstream water elevation only starts to affect the SFO discharge
capacity downstream of Entrance 4 at 50 m3 /s discharge as calculations in Table 3.3 show.
Spillway discharge capacity is
3/2
(3.1)
Q = CLe f f H0
where Q is total discharge, Le f f is the effective length of the spillway crest and H0 is
the total head on the spillway crest. The constant C is the dimensionless coefficient of
discharge derived as a multiplication of many factors, including a factor k2 which accounts
for downstream submergence. As shown in Table 3.3 all cases except for case shown gray
in the table, Entrance 4 at 50 m3 /s, the k2 factor is equal to one and in that case it is 0.9
yielding some effect on the SFO Entrance 4 discharge capacity.
70
Figure 3.38: Perspective views of the velocity magnitude at the water surface in the SFO
channel at discharge 50 m3 /s.
71
Figure 3.39: Water velocity magnitudes at cross section shown in Figure 3.36 inside the
SFO for discharges 20 m3 /s, 30 m3 /s, 40 m3 /s and 50 m3 /s.
72
Table 3.3: Calculations of the downstream water elevation on the SFO discharge capacity.
Hd /Ha is the ratio between the difference of critical water elevation on the SFO crest and
the water elevation downstream of the SFO crest (Hd ) and the water depth on the SFO
crest (H)
Critical water elevation Water elevation inside
Discharge
on the SFO crest
downstream of crest Hd /Ha k2
3
[m /s]
[m a.s.l.]
[m a.s.l.]
20
49.77
47.36
3.6
1
30
49.98
48.23
2.0
1
40
50.16
48.49
1.6
1
50
50.34
48.79
1.2
1
20
49.77
47.51
3.4
1
30
49.98
48.37
1.8
1
40
50.16
48.95
1.1
1
50
50.34
49.18
0.9
1
20
49.77
48.09
2.5
1
30
49.98
48.45
1.7
1
40
50.16
48.87
1.2
1
50
50.34
49.36
0.8
1
20
49.77
48.10
2.5
1
30
49.98
48.69
1.5
1
40
50.16
48.77
1.3
1
50
50.34
49.56
0.6
0.9
73
SFO Entrance 2
53
52
52
51
51
Elevation [m a.s.l.]
Elevation [m a.s.l.]
SFO Entrance 1
53
50
49
48
47
46
45
50
49
48
47
46
4 5 6
Station [m]
45
53
52
52
51
51
50
49
48
47
46
45
4 5 6
Station [m]
SFO Entrance 4
53
Elevation [m a.s.l.]
Elevation [m a.s.l.]
SFO Entrance 3
50
49
48
47
46
4 5 6
Station [m]
3
Q = 20 m /s
45
Q = 30 m /s
4 5 6
Station [m]
Q = 40 m /s
Q = 50 m /s
Figure 3.40: Water levels inside the SFO channel at longitudinal sections perpendicular
to the middle of each SFO entrance.
74
75
Figure 3.42: Water levels and velocity vectors at longitudinal sections in the SFO at
discharge 40 m3 /s.
3.3.3 SFO Summary
The flow characteristics inside the SFO conveyance channel are in general satisfactory
for the discharges tested. In general the SFO layout fulfils the design criteria set forth in
Section 2.3. The flow enters the side way channel, takes a swift 90 degree turn and carries
on without delay downstream in the channel. Flow velocities range from 1.5 to 4.7 m/s
in the conveyance channel where local abrupt changes in acceleration are only observed
downstream of the SFO entrances and at Station 25 where a drop in water elevation is
observed because of channel contraction. The SFO channel is smooth with no intrusions,
sharp corners, rough walls or locations where sediment or debris is likely to accumulate.
When water flows over a spillway crest the water can get aerated with bubbles forming in
the water. If bubbles are carried deep into a stilling basin or a downstream pool the water
76
3.4 Verification
can get supersaturated with dissolved gases, oxygen and nitrogen, when the gases inside
the bubbles are dissolved into the water by pressure (Pickett et al., 2004). Supersaturation of dissolved gases in water can cause gas bubble trauma in fish which can severely
traumatise the fish or even lead it to death (Fidler & Miller, 1994). The probability of
supersaturation of dissolved gases because of air entrainment in the SFO is low, although
air may get entrained into the water, the depth inside the channel and pressures are not
large enough to dissolve the gases into the water phase.
In general, the SFO should be able to carry the juvenile salmon safely and with out delay
downstream to the tailwater of the dam.
3.4 Verification
Solutions of numerical models can be dependent on the quality and size of the computational grid or mesh. To evaluate whether the mesh had been refined to the point where
no significant changes where observed in the character of the flow behaviour in the numerical model additional models with different mesh sizes were computed for the SFO
and reservoir models. The reservoir models were computed for conditions representing
Case A while the SFO models were computed for a discharge of 40 m3 /s. In Table 3.4
relevant parameters of the meshes tested are shown.
Table 3.4: Parameters of computational meshes used for verification of main mesh. The
parameters for the meshes used in the main test cases highlighted in gray.
Model
Number of nodes Number of elements Percentage of final mesh size
Reservoir
1930440
1866768
100
Reservoir
1715580
1657713
89
Reservoir
1582932
1528549
82
SFO
1206225
1153540
119
SFO
1010460
963380
100
SFO
587955
555264
58
SFO
332253
310141
33
Figure 3.43 shows velocity distribution at Section 20 in the reservoir model (location of
sections shown in Figure 3.12). As seen in the figure the velocity distribution is very
similar with only slight difference between smaller two meshes and the largest at Stations
40 and from Station 120 to Station 130. It can therefore be concluded that the mesh used,
M1930, is sufficient for evaluation of the approach flow characteristics.
77
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
M1583
20
40
60
80
Station [m]
100
M1716
M1930
120
140
Figure 3.43: Velocity distribution at Section 20 for different reservoir mesh sizes. M1583
= 1528549 nodes, M1716= 1715580 nodes and M1930 = 1930440 nodes.
Figure 3.44 shows water elevations at a cross section in the SFO model shown in Figure
3.36. The figure shows the water elevations are very similar for all mesh sizes especially
for meshes M588, M1010 and M1206. A small difference is observed at Station 20 for
Mesh M332. It can therefore be concluded that the mesh used for the evaluation, M1010,
is sufficient to show the character of the SFO conveyance channel.
Elevation [m a.s.l.]
50
49.5
49
48.5
48
47.5
47
46.5
46
0
10
15
20
25
30
35
40
45
50
55
60
Station [m]
M332
M588
M1010
M1206
Figure 3.44: Water elevations for different SFO mesh sizes at a cross section. M332 =
332253 nodes, M588 = 587955 nodes, M1010 = 1010460 nodes and M1206 = 1206225
nodes.
78
4 Conclusion
In this thesis, results from a physical model and a numerical model of Heiarln Reservoir are compared upstream of the proposed Surface Flow Outlet (SFO) structure. The
models are tested with regard to parameters related to fish passage in order to evaluate
the effectiveness of the SFO structure. The study also includes an investigation of the
SFO conveyance channel with a separate numerical model. The SFO conveyance channel
model is tested and evaluated with respect to flow conditions and other features which
may prove hazardous to the juvenile salmon as it is transported downstream to the tail
waters.
In general the overall flow characteristic of the reservoir models are in good agreement
and can be used to evaluate the effectiveness of the approach flow channel and the SFO
layout. A stagnant velocity zone forms immediately upstream of the spillway, the extent
of the stagnant zone is directly related to spillway discharge, reducing with increased
spillway discharge. Juvenile salmon entering the stagnant velocity zone immediately upstream of the spillway may loose track of the attraction flow causing delay on their way to
the ocean. The models differ most at the left approach bank where a current coming over
the bank intersects the main current in the approach flow channel at an steep angle forming
irregularities in the flow. The irregularities are not likely to affect the juvenile salmon as
the irregularities do not affect the attraction flow significantly. To prevent negative impact
during the operation of the power plant aforementioned abnormalities in the approach
flow should be considered.
The attraction flow towards the SFO is extensive reaching far upstream and should guide
the juvenile salmon in an safe and timely manner through the structures. Separation of
flow between the spillway and intake and SFO is consistent between the physical and the
numerical models. Results from a dye test conducted in the physical model and numerical
model results indicate the SFO transporting water from depths ranging from 1 m to 2.7 m
depending on project operation. The numerical model showed more distinct separation of
streamlines at greater depth than the physical model where distinguishing the separation
depth proved not trivial due to vortices scattering the dye vertically. The numerical model
result show a smooth acceleration towards the SFO, the flow accelerates 1 m/s per meter
over the last 2 m in front of the SFO. Maximum values in velocity magnitude of almost
3 m/s are observed at the SFO crests which should suffice to capture the juvenile salmon.
During the migration period 91% of the juvenile salmon are expected to enter the SFO
(Karadttir & Gudjonsson, 2012b).
79
4 Conclusion
The SFO model presents an estimation of the flow conditions inside the SFO conveyance
channel. Water elevations are at maximum downstream of the upstream most entrances,
Entrances 3 and 4, with water elevations higher than SFO crest elevation at discharges 40
m3 /s and 50 m3 /s. A distinct draw down is observed downstream of Entrance 1 where the
width of the channel decreases. Because the flow regime is subcritical the contraction in
the channel causes a draw down in water level. The flow conditions inside the conveyance
channel are satisfactory and show the SFO design providing means of safe and timely
passage of the juvenile salmon through the project.
The combination of numerical and physical models proves to be a valuable tool in the
evaluation of a hydraulic design. The models strengthen each other and with good comparison between the models comes the opportunity to look into features in the numerical
model which are hard or impossible to measure in the physical model. The good comparison also provides a calibrated tool that can be used after the physical model is no longer
available.
80
References
ANSYS (2011). ANSYS CFX Release 14.0 - Theory guide [Computer software manual].
Antonsson, T., & Gudjonsson, S. (2002). Variability in timing and characteristics of
atlantic salmon smolt in Icelandic rivers. Transactions of the American Fisheries
Society, 131, 643-655.
Bell, M. C. (1990). Fisheries handbook of engineering requirements and biological
criteria (3rd ed.). U.S. Army Corps of Engineers, North Pacific Division.
Bureau of Reclamation, H. I., & Group, L. S. (2011). WinADV - Version 2.028.
Chanson, H. (2004). The hydraulics of open channel flow: An introduction. Elsevier.
Coutant, C. C., & Whitney, R. R. (2000). Fish behavior in relation to passage through
hydropower turbines: A review. Transactions of the American Fisheries Society,
129(Iss.2), 351-380.
Fidler, L. E., & Miller, S. B. (1994, September). British Columbia water quality guidelines for dissolved gas supersaturation.
Goodwin, R. A. (2003). Hydrodynamics and juvenile salmon movement behaviour at
lower grandite dam: Decoding the relationship using 3-d space-time (CEL AGENT
IBM) simulation. Unpublished doctoral dissertation, Cornell University.
Goodwin, R. A., Nestler, J. M., Anderson, J. J., Weber, L. J., & Loucks, D. P. (2006).
Forecasting 3-d fish movement behavior using a Eulerian-Lagrangian-Agne Method
(ELAM). Ecological modeling, 192, 197-223.
Goodwin, R. A., Nestler, J. M., Weber, L., Lai, Y. G., & Loucks, D. P. (2001).
Ecologically sensitive hydraulic design for rivers:Lessons learned in coupled
modeling for improved fish passage. In Wetlands engineering and river restoration
proceedings of Wetlands Engineering and River Restoration Conference 2001.
Gudjonsson, S., & Johannesson, H. (2009). Laxaseii og hverflar virkjunum Neri
jrs (Tech. Rep.). Landsvirkjun, Veiimlastofnun.
Gudmundsson, A., Gardarsson, S. M., Tomasson, G. G., Gunnarsson, A., & Johannesson, H. (2012). Model investigation of a juvenile bypass system at Urridafoss
hydroelectric project in iceland. In Hydro 2012,innovative approaches to global
challenges, Bilbao, Spain. The international journal on hydropower and dams.
Gunnarsson, A., Gardarsson, S. M., Tomasson, G. G., & Johannesson, H. (2012). Stilling
basin length optimization for Hvammur hydro electric project. In Proceedings from
IAHR 2nd Europe Congress, Munich.
Gunnnarsson, A. (2012). Physical model investigation on the Hvammur hep spillway.
,M.Sc. thesis, Faculty of Civil and Environmental Engineering, University of Iceland.
81
References
Haro, A., Odeh, M., Noreika, J., & Castro-Santos, T. (1998). Effect of water acceleration
on downstream migratory behavior and passage of Atlantic salmon smolts and juvenile American shad at surface bypasses. Transactlons of the American Fisheries
Society, 127(1), 118-127.
Helwig, L., Katzenberger, G., & Stensby, D. (1999). A new juvenile fish bypass system
at bonneville dam. In Waterpower 99 - Hydros Future: Technology, markets, and
policy proceedings of Waterpower Conference 1999. ASCE.
Hirt, C. W., & Nichols, B. D. (1981). Volume of Fluid (VOF) method for the dynamics
of free boundaries. Journal of Computational Physics, 39, 201-215.
Hrafnsdottir, H., & Freysteinsson, S. (2009). jrs river - South Iceland, Hvammur and Urriafoss hydroelectric projects, ice jam evaluation (Tech. Rep. No. LV2009/127). Landsvirkjun.
Johannsson, M., Jonsson, B., Ornolfsdottir, E. B., Gudjonsson, S., & Magnusdottir, R.
(2002, September). Rannsknir lfrki jrsr og vera hennar vegna virkjana
nean Brfells. Veiimlastofnun. (Unni fyrir Landsvirkjun)
Johnson, G. E. (2006). Hydroacoustic evaluation of juvenile salmonid passage at the
Dalles dam sluiceway, 2005. Pacific Northwest National Laboratory.
Johnson, G. E. (2009). Smolt responses to hydrodynamic conditions in forebay flow nets
of surface flow outlets, 2007. Pacific Northwest National Laboratory.
Johnson, G. E., & Dauble, D. D. (2006). Surface flow outlets to protect juvenile
salmonids. Reviews in Fisheries Science, 14, 213-244.
Karadottir, O. R., & Gudjonsson, S. (2009). Nth-60 Neri jrs minnisbla, mlefni:
Fiskvegir vi Urriafossvirkjun. Landsvirkjun, Verks, Veiimlastofnun. (VERKNMER: 2007.0306)
Karadttir, O. R., & Gudjonsson, S. (2012a). Nth-60 Neri jrs minnisbla, mlefni:
Seiafleyta - string flgtta gngutma seia. Landsvirkjun, Verks, Veiimlastofnun. (VERKNMER: 11036-001)
Karadttir, O. R., & Gudjonsson, S. (2012b). Nth-60 Neri jrs minnisbla, mlefni:
Seiafleyta Urriafossvirkjun - lkanprfanir (Tech. Rep.). Landsvirkjun, Verks,
Veiimlastofnun. (VERKNMER: 11036-001)
Khan, L. A., Wicklein, E. A., Rashid, M., Ebner, L. L., & Richards, N. A. (2008). Case
study of and application of a computational fluid dynamics model to the forebay of
the Dalles Dam, Oregon. Journal of Hydraulic Engineering, 134(5), 509-519.
Landsvirkjun. (2010-2012). Engineering drawings issued by the designers during the design phase of the project. (Drawings might be altered from original). Dupplied by
Mannvit and Verks Engineering.
Lee, H., Lin, C.-L., & Weber, L. J. (2008, July 1). Application of a nonhydrostatic model
to flow in a free surface fish passage facility. Journal of Hydraulic Engineering,
134(7).
Marchall, E., Mather, M., Parrish, D., Allison, G., & McMenemy, J. (2011). Migration
delays caused by anthropogenic barriers: Modeling dams, temperature, and success
of migrating salmon smolts. Ecological Applications.
Meselhe, E. A., & Odgaard, A. J. (1998). 3d numerical flow model for fish diversion
studies at wanapum dam. Journal of Hydraulic Engineering, 124(12), 1203-1214.
82
References
NOAA. (2008, February). Anadromous salmonid passage facility design. Retrieved from
www.nwr.noaa.gov
NRCS. (2007, August). Technical supplement 14n, fish passage and screening design,
part 654 of the National Engineering Handbook.
Pfister, M., & Chanson, H. (2011). Scale effects in physical hydraulic engineering models.
Journal of Hydraulic Research, 49(3), 293-306.
Pickett, P. J., Rueda, H., & Herold, M. (2004, June). Total maximum daily load for
total dissolved gas in the Mid-Columbia River and Lake Roosevelt (Tech. Rep.).
Washington State Department of Ecology, U.S. Environmental Agency, Washington
State Department of Ecology. (Submittal Report)
Schilt, C. R. (2006). Review: Developing fish passage and protection at hydropower
dams. Applied Animal Behaviour Science.
Seear, P. J., Carmichael, S. N., Talbot, R., Taggart, J. B., Bron, J. E., & Sweeney,
G. E. (2010). Differential gene expression during smoltification of Atlantic salmon
(salmo salar l.): a first large-scale microarray study. Mar Biotechnol, 12, 126-140.
Sontek. (1997). Acoustic Doppler Velocimeter technical documentation v.4.0. San Diego,
USA.
Sweeney, C. E., Hall, R., Giorgi, A. E., Miller, M., & Johnson, G. E. (2007, December). Surface bypass program comprehensive review report. ENSR Corporation.
(Prepared for U.S. Army Corps of Engineers, Portland District)
Tomasson, G. G., Gardarsson, S. M., & Gunnarsson, A. (2012, December). Hvammur
HEP Lower jrs: Physical model investigation on the spillway and juvenile fish
passage. Final report. Landsvirkjun, University of Iceland, Reykjavk University.
Tomasson, G. G., Gardarsson, S. M., Gunnarsson, A., Petursson, G. S., & Gudmundsson,
A. (2012, December). Urriafoss HEP Lower jrs: Physical model investigation
on the spillway and juvenile fish passage. Final report. Landsvirkjun, University of
Iceland, Reykjavk University.
Weber, L. J., Goodwin, R. A., Li, S., Nestler, J. M., & Anderson, J. J. (2006). Application
of an Eulerian Lagrangian Agent Method(ELAM) to rank alternative designs of a
juvenile fish passage facility. Journal of Hydroinformatics, 08.4.
83